Comparing of the Maximum Likelihood (Ml) and the Least Squares (Ls) Methods in Terms of Variance Components for Unequal Numbers of Abservations in Subclasses

Abstract

In this study, two parameter estimators. Maximum Likelihood (ML) and Least Squere (LS) methods, have been compared in case of random and mixed model conditions with respect to the efficiency of the estimated parameters. According to results obtained, ML method should be preferred to LS method in the case of random and mixed models for unequal numbers of observation in subclasses.